Reproducibility in Science

👤Ondřej Mottl

Science School of Quantitative Ecology 2025

bit.ly/SSoQE

This presentation


😸 Code on GitHub

🖥️ Slides

Research cycle

This image was created by Scriberia for The Turing Way community and is used under a CC-BY licence

Reproducibility crisis

Baker, Monya. “1,500 scientists lift the lid on reproducibility.” Nature 533, no. 7604 (2016).

Evolution

This image was created by Scriberia for The Turing Way community and is used under a CC-BY licence

Discussion


What is Open Science to You?

03:00

Open Science

UNESCO (2022). Understanding open science — Factsheet — SC-PBS-STIP/2022/OST/1.

A better view

https://www.fosteropenscience.eu/

The Journey

This image was created by Scriberia for The Turing Way community and is used under a CC-BY licence

Reproducibility

Reproducibility ?


Uknown author

I was able to reproduce my results on my computer/machine

Can I consider my work reproducible?

makeameme.org

Reproducibility

This image was created by Scriberia for The Turing Way community and is used under a CC-BY licence

Time to change

This image was created by Scriberia for The Turing Way community and is used under a CC-BY licence

It is a spectrum

Roger D. Peng ,Reproducible Research in Computational Science.Science334,1226-1227(2011).DOI:10.1126/science.1213847

Discussion


How reproducible is your research?

What can you do to improve it?

05:00

It is a spectrum

Getting started with reproducibility in research!

PROJECTS

Making a paper (compendium)



Project-oriented structure

Each paper is a single project

This image was created by Scriberia for The Turing Way community and is used under a CC-BY licence

Project communication plan

Plan.io

Project structure

This image was created by Scriberia for The Turing Way community and is used under a CC-BY licence

Project structure

# .
# └─my_awesome_project/
#   ├─ Data/
#   ├─ Outputs/
#   ├─ R/
#   ├─ README.md
#   ├─ LICENSE
#   └─ [project name].Rproj

Project structure

# .
# └─my_awesome_project/
#   ├─ Data/
#   |   ├─ Input/
#   |   ├─ Processed/
#   |   └─ Temp/
#   ├─ Outputs/
#   |   ├─ Data/
#   |   ├─ Figures/
#   |   └─ Tables/
#   ├─ R/
#   |   ├─ ___Init_project___.R
#   |   ├─ 00_Config_file.R
#   |   ├─ 01_Data_processing/
#   |   ├─ 02_Main_analyses/
#   |   ├─ 03_Supplementary_analyses/
#   |   ├─ Functions/
#   |       └─ example_fc.R
#   ├─ README.md
#   ├─ LICENSE
#   └─ [project name].Rproj

Project structure

# .
# └─my_awesome_project/
#   ├─ Data/
#   ├─ Outputs/
#   ├─ R/
#   |   ├─ ___Init_project___.R
#   |   ├─ 00_Config_file.R
#   |   ├─ 00_Master.R
#   |   ├─ 01_Data_processing/
#   |   |   ├─ 01_Prepare_pollen_data.R
#   |   |   └─ 01_Download_terrain_data.R
#   |   ├─ 02_Main_analyses/
#   |   |   ├─ 01_Vegetation_history/
#   |   |   |   ├─ 01_Estimate_dissimilarty.R
#   |   |   |   └─ 02_Summarise_dissimilarity.R
#   |   |   ├─ 02_Rate_of_change/
#   |   |   |   ├─ 01_Roc_estimation.R
#   |   |   |   ├─ 02_Roc_interpolation.R
#   |   |   ├─ 03_Temporal_patters_of_groups/
#   |   |   |   ├─ 01_Define_groups.R
#   |   |   |   ├─ 02_Temporal_patterns_of_groups.R
#   |   |   └─ 04_Visualisation/
#   |   ├─ 03_Supplementary_analyses/
#   |   └─ Functions/
#   ├─ README.md
#   ├─ LICENSE
#   └─ [project name].Rproj

Discussion


  • What is your current project structure?
  • What benefits do you think it has?

03:00

RStudio Projects


R-studio is using Projects already by default

Reproducible Data Science

Self-containedness

Working Directory



@JennyBryan:

If the first line of your R script is

setwd("C:\Users\jenny\path\that\only\I\have")

I will come into your office and SET YOUR COMPUTER ON FIRE 🔥.

The {here} package is up for rescue!

Artwork by @allison_horst

Self-containedness

R needs to have the environment set up correctly to run the code!

This includes the packages used in the code.

Tutorials for BIOL202

Self-containedness




Will my code run on your machine?

Will my code run in 10 years?

The {renv} package will make sure it will!

Artwork by @allison_horst

Note on practical exercises

image by adiasarahma

Practical Exercise


  • Install the {renv} package
  • Make a new project and initialize {renv} with renv::init()
  • Create a snapshot with renv::snapshot()

15:00

Self-containedness

More advanced!

What if we share the whole Operating System as well?

Icon by icon-icons.com; artwork by @statworx

Data Science

Why R for Data Science?

R is specifically designed for data science:

  • Import data from various sources
  • Tidy data into consistent structure
  • Transform data for analysis
  • Visualize patterns and relationships
  • Model to understand and predict
  • Communicate results effectively

All wrapped in Programming for reproducibility!

R for Data Science by Hadley Wickham, Mine Çetinkaya-Rundel, and Garrett Grolemund

R for Data Science

R for Data Science - Free online at r4ds.hadley.nz

Discussion


05:00

{tidyverse}

My suggestion: Let’s embrace the {tidyverse}!

LinkeIn - Nicholas Bell, Ph.D.

Data wrangling

Centre for Statistics in Ecology, the Environment and Conservation (SEEC)

Data wrangling

create meme in R

Code Exercise

GOAL: Learn basic {tidyverse} principles and functions.


Materials:

15:00

Key Takeaways 🎯

Why Tidyverse?

  1. Readability: Code reads like English, making it easier to understand and maintain
  2. Consistency: All functions follow the same design principles
  3. Pipes: Chain operations together logically
  4. Data-first: All functions take data as the first argument (pipe-friendly)
  5. Tidy data: Encourage proper data structure for analysis

Immutability

Do not edit raw data!

Have a record of all changes to raw data!

Best practice:

  • Data/Input/Raw/
  • Data wrangling in code
  • Save processed data (Data/Processed/)

data_query_memes

Immutability




  1. Do not overwrite R objects!
    • make a new object instead
  2. Do not overwrite data columns!
    • make a new column instead

arcserve

Code Exercise

GOAL: Learn basic Immutability principles.


Materials:

15:00

The Immutability Checklist

✅ File Level:

✅ Object Level:

✅ Column Level:

Summary of Data Science

Use {tidyverse} !

Embrace Immutability !

🚀 Practical Outcomes:

  • Debugging: Can isolate exactly where problems occur
  • Peer Review: Reviewers can trace your analytical logic step-by-step
  • Future You: You’ll understand your analysis months later
  • Collaboration: Others can build on and verify your work
  • Extensions: Easy to modify analysis without starting over

Outro

SPROuT

The materials used in this presentation are further expanded in a greater detail in course Science Powered through Reproducibility, Openness, and Teamwork (SPRouT) taught at Charles University.

About me

Ondřej Mottl Assistant Professor at Charles University

Head of the 🧑‍💻 Laboratory of Quantitative Ecology